A generalist / roleplaying model merge based on Llama 3. Models are selected from my personal experience while using them.
I personally think this is an improvement over Stheno v3.2, considering the other models helped balance out its creativity and at the same time improving its logic.
Settings:
Instruct // Context Template: Llama-3-Instruct
Temperature: 1.4
min_p: 0.1
Merging seems to be a black box magic though? In my personal experience merging multiple models from different datasets / data works better than combining them all in one.
Values chosen are from long-running personal experimentation since Llama-2 Merging Era. I have tweaked them to fit this recipe.
Mergekit Config
models:
- model: meta-llama/Meta-Llama-3-8B-Instruct
- model: crestf411/L3-8B-sunfall-v0.1 # Another RP Model trained on... stuff
parameters:
density: 0.4
weight: 0.25
- model: Hastagaras/Jamet-8B-L3-MK1 - # Another RP / Storytelling Model
parameters:
density: 0.5
weight: 0.3
- model: maldv/badger-iota-llama-3-8b #Megamerge - Helps with General Knowledge
parameters:
density: 0.6
weight: 0.35
- model: Sao10K/Stheno-3.2-Beta # This is Stheno v3.2's Initial Name
parameters:
density: 0.7
weight: 0.4
merge_method: ties
base_model: meta-llama/Meta-Llama-3-8B-Instruct
parameters:
int8_mask: true
rescale: true
normalize: false
dtype: bfloat16